From: Chris Metcalf <cmetcalf@tilera.com> To: Gilad Ben-Yossef <gilad@benyossef.com> Cc: <linux-kernel@vger.kernel.org>, Peter Zijlstra <a.p.zijlstra@chello.nl>, Frederic Weisbecker <fweisbec@gmail.com>, Russell King <linux@arm.linux.org.uk>, <linux-mm@kvack.org>, Christoph Lameter <cl@linux-foundation.org>, Pekka Enberg <penberg@kernel.org>, Matt Mackall <mpm@selenic.com> Subject: Re: [PATCH 0/5] Reduce cross CPU IPI interference Date: Wed, 28 Sep 2011 09:00:41 -0400 [thread overview] Message-ID: <4E831A79.1030402@tilera.com> (raw) In-Reply-To: <1316940890-24138-1-git-send-email-gilad@benyossef.com> On 9/25/2011 4:54 AM, Gilad Ben-Yossef wrote: > We have lots of infrastructure in place to partition a multi-core system such > that we have a group of CPUs that are dedicated to specific task: cgroups, > scheduler and interrupt affinity and cpuisol boot parameter. Still, kernel > code will some time interrupt all CPUs in the system via IPIs for various > needs. These IPIs are useful and cannot be avoided altogether, but in certain > cases it is possible to interrupt only specific CPUs that have useful work to > do and not the entire system. > > This patch set, inspired by discussions with Peter Zijlstra and Frederic > Weisbecker when testing the nohz task patch set, is a first stab at trying to > explore doing this by locating the places where such global IPI calls are > being made and turning a global IPI into an IPI for a specific group of CPUs. > The purpose of the patch set is to get feedback if this is the right way to > go for dealing with this issue and indeed, if the issue is even worth dealing > with at all. I strongly concur with your motivation in looking for and removing sources of unnecessary cross-cpu interrupts. We have some code in our tree (not yet returned to the community) that tries to deal with some sources of interrupt jitter on tiles that are running isolcpu and want to be 100% in userspace. > This first version creates an on_each_cpu_mask infrastructure API (derived from > existing arch specific versions in Tile and Arm) and uses it to turn two global > IPI invocation to per CPU group invocations. The global version looks fine; I would probably make on_each_cpu() an inline in the !SMP case now that you are (correctly, I suspect) disabling interrupts when calling the function. > The patch is against 3.1-rc4 and was compiled for x86 and arm in both UP and > SMP mode (I could not get Tile to build, regardless of this patch) Yes, our gcc changes are still being prepped for return to the community, so unless you want to grab the source code on the http://www.tilera.com/scm website, you won't have tile support in gcc yet. (binutils has been returned, so gcc is next up.) -- Chris Metcalf, Tilera Corp. http://www.tilera.com
WARNING: multiple messages have this Message-ID (diff)
From: Chris Metcalf <cmetcalf@tilera.com> To: Gilad Ben-Yossef <gilad@benyossef.com> Cc: linux-kernel@vger.kernel.org, Peter Zijlstra <a.p.zijlstra@chello.nl>, Frederic Weisbecker <fweisbec@gmail.com>, Russell King <linux@arm.linux.org.uk>, linux-mm@kvack.org, Christoph Lameter <cl@linux-foundation.org>, Pekka Enberg <penberg@kernel.org>, Matt Mackall <mpm@selenic.com> Subject: Re: [PATCH 0/5] Reduce cross CPU IPI interference Date: Wed, 28 Sep 2011 09:00:41 -0400 [thread overview] Message-ID: <4E831A79.1030402@tilera.com> (raw) In-Reply-To: <1316940890-24138-1-git-send-email-gilad@benyossef.com> On 9/25/2011 4:54 AM, Gilad Ben-Yossef wrote: > We have lots of infrastructure in place to partition a multi-core system such > that we have a group of CPUs that are dedicated to specific task: cgroups, > scheduler and interrupt affinity and cpuisol boot parameter. Still, kernel > code will some time interrupt all CPUs in the system via IPIs for various > needs. These IPIs are useful and cannot be avoided altogether, but in certain > cases it is possible to interrupt only specific CPUs that have useful work to > do and not the entire system. > > This patch set, inspired by discussions with Peter Zijlstra and Frederic > Weisbecker when testing the nohz task patch set, is a first stab at trying to > explore doing this by locating the places where such global IPI calls are > being made and turning a global IPI into an IPI for a specific group of CPUs. > The purpose of the patch set is to get feedback if this is the right way to > go for dealing with this issue and indeed, if the issue is even worth dealing > with at all. I strongly concur with your motivation in looking for and removing sources of unnecessary cross-cpu interrupts. We have some code in our tree (not yet returned to the community) that tries to deal with some sources of interrupt jitter on tiles that are running isolcpu and want to be 100% in userspace. > This first version creates an on_each_cpu_mask infrastructure API (derived from > existing arch specific versions in Tile and Arm) and uses it to turn two global > IPI invocation to per CPU group invocations. The global version looks fine; I would probably make on_each_cpu() an inline in the !SMP case now that you are (correctly, I suspect) disabling interrupts when calling the function. > The patch is against 3.1-rc4 and was compiled for x86 and arm in both UP and > SMP mode (I could not get Tile to build, regardless of this patch) Yes, our gcc changes are still being prepped for return to the community, so unless you want to grab the source code on the http://www.tilera.com/scm website, you won't have tile support in gcc yet. (binutils has been returned, so gcc is next up.) -- Chris Metcalf, Tilera Corp. http://www.tilera.com -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux-mm.org/ . Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/ Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
next prev parent reply other threads:[~2011-09-28 13:00 UTC|newest] Thread overview: 66+ messages / expand[flat|nested] mbox.gz Atom feed top 2011-09-25 8:54 [PATCH 0/5] Reduce cross CPU IPI interference Gilad Ben-Yossef 2011-09-25 8:54 ` Gilad Ben-Yossef 2011-09-25 8:54 ` [PATCH 1/5] smp: Introduce a generic on_each_cpu_mask function Gilad Ben-Yossef 2011-09-25 8:54 ` Gilad Ben-Yossef 2011-09-25 11:37 ` Sasha Levin 2011-09-25 11:37 ` Sasha Levin 2011-09-26 8:48 ` Peter Zijlstra 2011-09-26 8:48 ` Peter Zijlstra 2011-09-25 8:54 ` [PATCH 2/5] arm: Move arm over to generic on_each_cpu_mask Gilad Ben-Yossef 2011-09-25 8:54 ` Gilad Ben-Yossef 2011-09-25 8:54 ` [PATCH 3/5] tile: Move tile to use " Gilad Ben-Yossef 2011-09-25 8:54 ` Gilad Ben-Yossef 2011-09-25 8:54 ` [PATCH 4/5] mm: Only IPI CPUs to drain local pages if they exist Gilad Ben-Yossef 2011-09-25 8:54 ` Gilad Ben-Yossef 2011-09-26 1:52 ` Shaohua Li 2011-09-26 1:52 ` Shaohua Li 2011-09-26 6:47 ` Gilad Ben-Yossef 2011-09-26 6:47 ` Gilad Ben-Yossef 2011-09-26 15:24 ` Christoph Lameter 2011-09-26 15:24 ` Christoph Lameter 2011-09-27 7:27 ` Gilad Ben-Yossef 2011-09-27 7:27 ` Gilad Ben-Yossef 2011-09-27 16:13 ` Christoph Lameter 2011-09-27 16:13 ` Christoph Lameter 2011-09-26 7:28 ` Peter Zijlstra 2011-09-26 7:28 ` Peter Zijlstra 2011-09-26 8:39 ` Gilad Ben-Yossef 2011-09-26 8:39 ` Gilad Ben-Yossef 2011-09-26 7:29 ` Peter Zijlstra 2011-09-26 7:29 ` Peter Zijlstra 2011-09-25 8:54 ` [PATCH 5/5] slub: Only IPI CPUs that have per cpu obj to flush Gilad Ben-Yossef 2011-09-25 8:54 ` Gilad Ben-Yossef 2011-09-26 6:54 ` Pekka Enberg 2011-09-26 6:54 ` Pekka Enberg 2011-09-26 7:36 ` Peter Zijlstra 2011-09-26 7:36 ` Peter Zijlstra 2011-09-26 8:07 ` Gilad Ben-Yossef 2011-09-26 8:07 ` Gilad Ben-Yossef 2011-09-26 10:03 ` Pekka Enberg 2011-09-26 10:03 ` Pekka Enberg 2011-09-26 8:10 ` Gilad Ben-Yossef 2011-09-26 8:10 ` Gilad Ben-Yossef 2011-09-26 7:33 ` Peter Zijlstra 2011-09-26 7:33 ` Peter Zijlstra 2011-09-26 8:35 ` Gilad Ben-Yossef 2011-09-26 8:35 ` Gilad Ben-Yossef 2011-09-26 9:28 ` Pekka Enberg 2011-09-26 9:28 ` Pekka Enberg 2011-09-26 9:45 ` Peter Zijlstra 2011-09-26 9:45 ` Peter Zijlstra 2011-09-26 12:05 ` Gilad Ben-Yossef 2011-09-26 12:05 ` Gilad Ben-Yossef 2011-09-26 13:49 ` Gilad Ben-Yossef 2011-09-26 13:49 ` Gilad Ben-Yossef 2011-09-26 7:20 ` [PATCH 0/5] Reduce cross CPU IPI interference Peter Zijlstra 2011-09-26 7:20 ` Peter Zijlstra 2011-09-26 8:43 ` Gilad Ben-Yossef 2011-09-26 8:43 ` Gilad Ben-Yossef 2011-09-26 8:46 ` Peter Zijlstra 2011-09-26 8:46 ` Peter Zijlstra 2011-09-28 13:00 ` Chris Metcalf [this message] 2011-09-28 13:00 ` Chris Metcalf 2011-10-02 8:44 ` Gilad Ben-Yossef 2011-10-02 8:44 ` Gilad Ben-Yossef 2011-10-02 14:58 ` Chris Metcalf 2011-10-02 14:58 ` Chris Metcalf
Reply instructions: You may reply publicly to this message via plain-text email using any one of the following methods: * Save the following mbox file, import it into your mail client, and reply-to-all from there: mbox Avoid top-posting and favor interleaved quoting: https://en.wikipedia.org/wiki/Posting_style#Interleaved_style * Reply using the --to, --cc, and --in-reply-to switches of git-send-email(1): git send-email \ --in-reply-to=4E831A79.1030402@tilera.com \ --to=cmetcalf@tilera.com \ --cc=a.p.zijlstra@chello.nl \ --cc=cl@linux-foundation.org \ --cc=fweisbec@gmail.com \ --cc=gilad@benyossef.com \ --cc=linux-kernel@vger.kernel.org \ --cc=linux-mm@kvack.org \ --cc=linux@arm.linux.org.uk \ --cc=mpm@selenic.com \ --cc=penberg@kernel.org \ /path/to/YOUR_REPLY https://kernel.org/pub/software/scm/git/docs/git-send-email.html * If your mail client supports setting the In-Reply-To header via mailto: links, try the mailto: linkBe sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.